[idn] draft-klensin-dns-role-00.txt and history

["A.M.Rutkowski" <amr@netmagic.com>]

Your paper on DNS (draft-klensin-dns-role-00.txt) is clearly one of the
most thoughtful and imaginative commentaries on this subject, and it
has usefully sparked a meta level effort among many of us to consider
how to evolve the DNS in significant ways.

While flying back from the IETF, however, one historical remark was
troublesome enough to produce this note.  At the risk of seeming
self-serving, I'd like to raise a few counter-points relevant to
the following statements - in particular the remark about "not
obviously benefited the Internet as a whole."

     Convenience of typing, and the desire to make domain
     names out of easily-remembered product names, has led to
     a flattening of the DNS, with many people now perceiving
     that second-level names under COM (or in some countries,
     second- or third-level names under the relevant ccTLD)
     are all that is meaningful (this perception has been
     reinforced by some domain name registrars who have been
     anxious to "sell" additional names).  And, of course, the
     perception that one needs a top-level domain per product,
     rather than a (usually organizational) collection of
     network resources has led to a rapid acceleration in the
     number of names being registered, a phenonenum that has
     clearly benefited registrars charging on a per-name
     basis, "cybersquatters", and others in the business of
     "selling" names, but has not obviously benefitted the
     Internet as a whole.


The available evidence suggests the contrary, and that indeed,
the Internet as a whole has incurred profound benefits from these
de facto implementations.  If it were just a matter of historical
perspective, I wouldn't have bothered writing this note.
However, it goes to a more significant underlying difficulty
going forward.  In making choices, how do we decide "what
benefits the Internet" or not, and on the basis of what metrics
or other evidence?

One of the most attractive attributes of the Internet has been its
emphasis on user-driven developments, rather than unilateral actions
by committees of engineers.  Indeed, one of the stark comparisons of
alternative approaches involves network object naming.

Compare, for example, the success of Internet DNS versus the OSI domain
structure.  The latter was elegantly engineered in standards committees
and regulated through copious rules and registration processes developed by
government agencies and trademark attorneys.  I can remember sitting in
some of the hundreds of committee meetings and when objections were
raised about hierarchical or expression complexities, hearing remarks
like "we will define what the user gets."

The OSI domain structure even had the playing field tilted in its
direction through denominations of "de jure" standards and mandated
government procurement specifications.  Yet the OSI DNS failed
miserably in the marketplace in the face of the more freeform,
simple, and adaptable "de facto" Internet DNS.

The failures were not only in the OSI world.  A great many Internet DNS
country code TLDs - generally those managed like OSI domains - also
failed badly.  Users potentially using those domains simply chose
available alternatives, rather than dealing with the imposed structure,
bureaucratic constraints, unresponsiveness, long delays, and frequently
very high costs, despite their administration by non-profit organizations.

In 1994, at the NSF-IEEE workshop panel on COM domain administration,
serious consideration was given to emulating OSI domain registration
processes, but ultimately (wisely) rejected.  That decision, coupled
with a decision make several years earlier to privatize administration
of the COM, NET, and ORG namespace through an entrepreneurial
for-profit organization, resulted in an explosion in the namespace
that approximated the demand by users for whom the money involved was
trivial.

The result profoundly altered the dynamics of the Internet's growth and
success on a massive public scale.  It allowed almost anyone to easily
and quickly acquire a name that met their needs to instantiate multiple
corporate, product, brand, or personal persona in the form of web or
email services.  This had a ripple effect that produced an enormous
demand for products and services that collectively brought hundreds of
billions of dollars into the economy.  This in turn attracted
innovators and venture capital worldwide into the Internet domain an a
scale never seen before.  It would be ludicrous to claim this was all
due to the unfettered availability of COM domain names, but that
availability was patently a significant factor.

Ironically, even the inherent constraints within the flat space
ultimately proved beneficial by spawning creativity by marketing
wordsmiths to produce an entire new lexicon unique to the Internet.
It's worth noting that users seem to have no problem adapting to flat
namespaces - and indeed seem to prefer them - as exemplified by the
Internet's greatest namespace success - AOL Internet Messenger with
135 million registered names and still expanding.

So the bottom line is that although the factors you raise have produced
a set of problems whose solutions are usefully explored by the paper,
to assert that these factors have provided no benefits to the Internet
as a whole, seems rather off the mark.  Perhaps more importantly going
forward, some mechanisms will be needed to accommodate the inevitable
tensions among what the engineering community might seek, what the network
can support, and what ordinary end users and the marketplace want.

with best regards,
tony